How to model the infrastructure for the Rails application with processor/memory intensive tasks (using EC2)?

view full story

http://serverfault.com – I need to model a Rails application that does video processing using Amazon AWS resources (EC2). Application allows users to convert videos. Videos are uploaded by users. Right now it is running on one instance, but I see the need to scale up / change things. Question: Does it make sense to create two instances: 1) Application front-end. Handles uploads, database stuff. 2) Video processing instance. Something high-memory, that is getting hit by the application N1. Will it be more efficient? Is there a way to have instance N2 working not constantly, but only when it is needed to process (HowTos)